TelevisaUnivision’s Post-Cookie Targeting Shows Promise for Seller-Defined Audiences

With Google PPS and SDA, the media company could fully monetize its whole audience

Did you know brand recall increases when viewers see an ad on TV AND streaming? Download "A Practical Playbook for Multiscreen TV" to learn more. Sponsored by EffecTV.

As the industry scrambles to find replacements for cookies, a lack of scale has been a recurring issue for buyers transacting on publishers’ first-party data.

Among a litany of third-party cookie alternatives, seller-defined audiences (SDA) promise publishers control of what signals they put into the bidstream. And unlike other solutions that rely on first-party data, SDA guarantees scale in that it’s designed to be transacted in the open exchange, whereas first-party data typically lends itself to individual direct deals between publishers and buyers.

What SDA has been lacking is data proving its effectiveness.

TelevisaUnivision has been using Google Ad Manager’s publisher provided signals (PPS), the tool that facilitates SDA, and early results show preliminary evidence that the paradigm can help publishers more fully monetize their audiences.

The media company found on average 1.24 audience categories per impression across its U.S. inventory, meaning it can categorize the
majority of its impressions into at least one audience cohort across properties, ultimately fully monetizing its audience with PPS. These signals are still in beta and only available to transact via DV360 and Google Ads, and not yet available for other third-party ad-tech vendors.

Time is ticking on Chrome’s deprecation of third-party cookies, and the industry is grasping at any evidence for which alternative audience-identifying tech will yield the most effective results. While SDA solves for scale—and can help publishers identify findings from their audiences—evidence that it increases CPMs or ad effectiveness for advertisers is still holding back wider adoption.

Alleviating SDA concerns

TelevisaUnivision’s findings help alleviate some concerns about SDA, a taxonomy of cohorts developed by trade body the Interactive Advertising Bureau (IAB), which helps publishers and buyers communicate contextual signals in the same language. (Google’s PPS are based on this taxonomy.)

One concern is that an industry-developed set of standards won’t effectively and completely describe a publisher’s distinctive audience scope. TelevisaUnivision’s findings show that buyers can use PPS without leaving publisher impressions on the table.

With PPS, we’re transacting on an additional basis and driving some additional spend.

Bethany Hillman, vp, data and advertising operations, TelevisaUnivision

Another qualm publishers have expressed about SDA is that they risk passing valuable data into the bidstream.

This was particularly vexing for TelevisaUnivision, which didn’t want to classify audiences with data it was already offering exclusively to advertisers via direct deals, said Bethany Hillman, vice president of data and advertising operations at TelevisaUnivision.

“[We wanted] to have that signal but not cannibalize our direct sales,” Hillman said, noting the test of PPS has found that this is possible. “[With PPS, we’re] transacting on an additional basis and driving some additional spend our way.”

TelevisaUnivision was able to enjoy the best of both the open exchange and direct deals by working with ad-tech firm Permutive to categorize its data, only offering signals to the bidstream that it was not currently offering in its direct deals. It still managed to activate its audiences against 188 categories of the IAB’s more than 1,600 audience segments.

While the results show how PPS can monetize publishers’ inventory, they don’t show if the cookie alternative can meaningfully increase publishers’ revenue. Hillman said Google has still not made data available showing how PPS incrementally increases CPMs. The companies wouldn’t share more specific details from their tests.

Mapping cohorts

TelevisaUnivision works with Permutive to package its data using the tech vendor’s proprietary audience cohorts. Once the audiences are in Google Ad Manager, Permutive can use Google’s mapping features to connect the cohorts to IAB’s Audience Taxonomy and then ultimately send these groupings into the bidstream. While a publisher does not need to work with an ad-tech partner like Permutive to use PPS, it can make the process less laborious, Hillman said.

In deciding how to collate TelevisaUnivision’s data into the PPS groupings, Permutive chose attributes that were not already present in direct deals, and also logical, said Michael Ogunjobi, head of publisher strategy at Permutive.

“If it makes sense, we map to it and if it doesn’t, we step away,” Ogunjobi said “It’s not likely every single impression has to generate an IAB category.”

The rate of 1.24 audience categories per impression does not mean that every person who visits TelevisaUnivison’s media properties perfectly maps to a category, Hillman said. Some users may map into 15 categories, while others, none at all.

New findings about audiences

Grouping audiences into the IAB’s categories, and not the contextual attributes TelevisaUnivision would naturally use, has revealed new findings, Hillman said. The media company realized a lot of its audience qualified as literature fans, a cohort it did not previously recognize. This can impact how inventory is sold, which advertisers can use to target and potentially inform editorial decisions, Hillman said.  

Another value of using PPS is enriching open exchange data that has become increasingly opaque as cookies are deprecated in more environments, Hillman said.

But even if publishers start trusting the viability of SDA solutions like PPS, advertisers must also buy in, which is still one of the protocol’s biggest hurdles to adoption, said Ratko Vidakovic, founder of ad-tech consultancy AdProfs.

“From an advertiser’s perspective, there is still the question of whether these categories assigned to impressions are accurate,” Vidakovic said. “Having one or more categories per impression is bad if the categories being assigned are applied in a fast and loose manner.”

Enjoying Adweek's Content? Register for More Access!